Con: Should we abandon the use of the MDRD equation in favour of the CKD-EPI equation?

Transcription

1 POLAR VIEWS IN NEPHROLOGY 2. Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150: Botev R, Mallie JP, Wetzels JF et al. The Clinician and Estimation of Glomerular Filtration Rate by Creatinine-based Formulas: Current Limitations and Quo Vadis. Clin J Am Soc Nephrol 2011; 6: Delanaye P, Schaeffner E, Ebert N et al. Normal reference values for glomerular filtration rate: what do we really know? Nephrol Dial Transplant 2012; 27: Glassock RJ, Rule AD. The implications of anatomical and functional changes of the aging kidney: with an emphasis on the glomeruli. Kidney Int 2012; 82: K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002; 39: S van den Brand JA, van Boekel GA, Willems HL et al. Introduction of the equation to estimate glomerular filtration rate in a Caucasian population. Nephrol Dial Transplant 2011; 26: Levey AS, de Jong PE, Coresh J et al. The definition, classification, and prognosis of chronic kidney disease: a KDIGO Controversies Conference report. Kidney Int 2011; 80: Nephrol Dial Transplant (2013) 28: doi: /ndt/gft Matsushita K, Mahmoodi BK, Woodward M et al. Comparison of risk prediction using the equation and the study equation for estimated glomerular filtration rate. JAMA 2012; 307: El-Nahas M. Cardio-Kidney-Damage: a unifying concept. Kidney Int 2010; 78: Bjork J, Jones I, Nyman U et al. Validation of the Lund-Malmo, Chronic Kidney Disease Epidemiology () and Modification of Diet in Renal Disease () equations to estimate glomerular filtration rate in a large Swedish clinical population. Scand J Urol Nephrol 2012; 46: Buron F, Hadj-Aissa A, Dubourg L et al. Estimating glomerular filtration rate in kidney transplant recipients: performance over time of four creatinine-based formulas. Transplantation 2011; 92: Nyman U, Grubb A, Sterner G et al. The equations to estimate GFR. Validation in the Swedish Lund- Malmo Study cohort. Scand J Clin Lab Invest 2011; 71: Iliadis F, Didangelos T, Ntemka A et al. Glomerular filtration rate estimation in patients with type 2 diabetes: creatinine- or cystatin C-based equations? Diabetologia 2011; 54: Received for publication: ; Accepted in revised form: Con: Should we abandon the use of the equation in favour of the equation? Pierre Delanaye 1, Hans Pottel 2 and Rossini Botev 3 Correspondence and offprint requests to: Pierre Delanaye; pierre_delanaye@yahoo.fr 1 Department of Nephrology-Dialysis-Transplantation, University of Liège, CHU Sart Tilman, Liège, Belgium, 2 Interdisciplinary Research Center, University of Leuven, Kortrijk, Belgium and 3 Department of Nephrology, Hawaii Permanente Medical Group, Honolulu, HI, USA INTRODUCTION The best overall index of renal function is the glomerular filtration rate (GFR) [1]. Since measuring the GFR can be cumbersome and costly, estimation of GFR is essential for the diagnosis and evaluation of chronic kidney disease (CKD), defined as kidney damage or GFR <60 ml/min/ 1.73 m 2 for 3 months and staged by levels of GFR according to the NKF KDOQI (The National Kidney Foundation Kidney Disease Outcomes Quality Initiative) guidelines published in 2002 [2]. These now decade-old guidelines are under consideration for revision by the KDIGO (Kidney Disease: Improving Global Outcomes) CKD Work Group [3]. Several authors have proposed in the past different equations to estimate GFR based on The Author Published by Oxford University Press on Downloaded from behalf of ERA-EDTA. All rights reserved. 1396

2 serum creatinine concentration, the latter being determined by the individual s muscular mass in steady state, and thus by age, gender and ethnicity [4 8]. Among these equations, the one proposed by Cockcroft and Gault in 1976 [4] was doubtless the most popular for many years because of its simplicity and ease to calculate at the patient s bedside. However, this equation is an estimation of the creatinine clearance, not of the GFR, and was particularly imprecise, notably in CKD patients. In 1999, an equation derived from measured GFR (mgfr) was developed in the Modification of Diet in Renal Disease () Study. This equation included the serum creatinine concentration and six other variables, subsequently abbreviated to four variables [5, 6]. This equation has been endorsed by the KDOQI guidelines and is currently used by most clinical laboratories [2]. Several limitations of the study equation were elaborated on in the literature [9 12], the major one being the systematic underestimation of mgfr >60 ml/min/1.73 m 2, which essentially translates in to overestimation of the CKD prevalence. A new equation, the Epidemiology Collaboration () equation, was proposed in 2009 for better estimation of GFR levels >60 ml/min/1.73 m 2 [7]. ASSESSING THE PERFORMANCE OF EQUATIONS (OR WHY WE NEED PRECISION) Four main statistical tools are used when performance of equations must be evaluated: bias, precision, accuracy and CKD classification. Bias is defined as the mean (or median) difference between the estimated GFR (egfr) and mgfr [13, 14]. It is a conceptualization of the systematic error. Because this error is systematic, it is relatively easy to assess. For example, if a particular equation systematically underestimates the mgfr by 5 or 10 ml/min/1.73 m 2, it is not difficult for the clinicians to interpret the egfr result for a given patient. Conversely, when considering the egfr results in population studies, such a systematic error will be misleading in terms of CKD prevalence or related mortality and morbidity-associated risk [15, 16]. This was specifically observed with the equation, which by underestimating the mean mgfr, overestimates the CKD stage 3 prevalence in the general population [7, 15 18]. Precision is a concept that is frequently forgotten by most authors but it is, however, fundamental. Precision is usually expressed as standard deviation (SD) or interquartile range (IQR) of the bias, and thus represents the spread of 68% (assuming normal distribution) or 50% of the values around the bias, respectively. It conceptualizes the random error, which is much more difficult to assess and correct than the bias. The conceptualization is easier for the clinician to understand when precision is expressed as SD instead of IQR, because the former permits the estimation of the 95% range of egfr values. Contrary to the bias, this type of error will have a low impact on the results in a population, e.g. the CKD prevalence. However, such an error could be misleading for the clinicians when interpreting an individual's egfr result. In a patient with an mgfr of 60 ml/min/1.73 m 2, an equation with a perfect bias (0 ml/min/1.73 m 2 ) but a global precision (SD) of 10 ml/min/1.73 m 2, would provide 95% probability so that the egfr value is between 40 and 80 ml/min/1.73 m 2, something the clinicians would not be aware of when an egfr value is routinely reported by a clinical laboratory. Accuracy combines bias and precision and it is easy to understand when presented as the percentage of egfrs within a defined range of their respective mgfrs. Root mean square error also combines bias and precision but its conceptualization is more difficult to interpret in practice by clinicians. In the KDOQI guidelines, a range of ± 30% with 90% of egfrs in this range, was cited as a useful measure of accuracy [2]. As practicing clinicians, we have to question if such a goal is really sufficient for important clinical decisions, as, for example, whether a potential candidate could be a living kidney donor. In such situations, an accuracy of ± 15% could be more useful [9, 14]. CKD classification is one of the most important statistical tools for the clinicians when evaluating the performance of an equation on an individual level [9, 19]. It is related to the percentage of patients correctly classified by egfr into the different CKD stages in comparison with the confirmatory test of mgfr [20]. Unfortunately, this statistical tool is not always used appropriately [9]. For example, some authors [7, 21] consider the egfr as a reference, while many others correctly used the mgfr as a reference for defining the five CKD stages when assessing an equation s performance [19, 22 32]. An equation with high correct CKD classification would decrease the need for determining the mgfr and provide great confidence to the clinicians to implement an appropriate plan of action according to the individual's egfr result. THE EQUATION: WHAT IS (REALLY) NEW? The relationship between GFR and serum creatinine is different in healthy subjects and CKD patients [33, 34]. This physiological fact explains why the equation, developed exclusively from a CKD population, underestimates the high GFR levels. Because the authors of the equation included a significant proportion of subjects with a normal GFR, the serum creatinine variable was modelled as a spline with sex-specific knots (0.7 and 0.9 mg/dl for women and men, respectively) and different exponents were applied to serum creatinine according to its level [7]. As expected, the performance of the new equation was globally better than the one. For example, in the external validation data set, the median bias (egfr-mgfr), precision (IQR) and accuracy ± 30% for the and equations were: 5.5 versus 2.5 ml/min/1.73 m 2, 18.3 versus 16.6 ml/min/1.73 m 2 and 81 versus 84%, respectively. However, it must be underlined that the improved performance of the equation was essentially due to a lower bias [3 ml/min/1.73 m 2 (55%) reduction], while precision remained comparable POLAR VIEWS IN NEPHROLOGY 1397 Estimating GFR using the

3 POLAR VIEWS IN NEPHROLOGY [1.7 ml/min.1.73 m 2 (9%) reduction] and suboptimal. Accuracy ± 30% was essentially unchanged [3.5% (4%) increase]. As expected too, for GFR >60 ml/min/1.73 m 2 the equation s bias was better [7 ml/min/1.73 m 2 (67%) reduction], but meanwhile its precision declined by 12.9 ml/ min/1.73 m 2 (114%), 11.3 versus 24.2 ml/min/1.73 m 2 for mgfr 60 and >60 ml/min/1.73 m 2, respectively, and was not significantly different compared with the equation [1.5 ml/min/1.73 m 2 (6%) reduction] [7]. How can we explain this absence of a better precision in the new equation? Two main explanations could be advanced. First, the precision of an estimator will strongly depend on the precision of measuring serum creatinine (as a main biological variable) and GFR. In the vast majority of studies used as development data set for the equation, serum creatinine has been measured by the Jaffe methods which are largely less precise, even if isotope dilution mass spectrometry traceable, than the enzymatic methods. This imprecision impacts the global accuracy of the new equation [35 37]. Hence, it is not surprising that the best accuracy of creatinine-based equations is seen in studies with serum creatinine measured by enzymatic assay in paediatric [38] or adult [14, 39] studies. In the development data set, only urinary clearances of subcutaneously injected iothalamate were performed to measure GFR. In this measurement, precision is not the best [40] and, incidentally, the clearance of iohexol has been shown to have a better precision and reproducibility [41 43]. Secondly, the relative lack of precision of the equation is probably due to the studied population [7]. Certainly, the population sample is impressive as the new equation has been built from a development data set including 5504 subjects (with an internal validation data set of 2750 subjects) and subsequently tested in an external validation data set of 3896 subjects. Although the cohort represented a relatively homogeneous population (CKD patients without diabetes), the one included a more heterogeneous population with potentially different GFR-serum creatinine relationships [33, 34]. After its introduction, the equation was studied in various populations. A PubMed database (last accessed 18 June 2012) search for GFR, and in adults with a minimum of 50 mgfrs (estimates from smaller studies can be unreliable) and provided data for ± 30% accuracy recovered 26 publications [7, 21, 26, 32, 39, 44 64]. The results for accuracy, bias and precision and their respective calculated weighted average values are presented in Table 1. The equation had a slightly better weighted average for ±30% accuracy of 1.8%, mean/median bias of 3.5/0.1 ml/min/1.73 m 2, respectively, and precision of 1.1 ml/min/1.73 m 2 compared with the one. These differences are not clinically significant with the exception of a better mean bias of 3.5 ml/ min/1.73 m 2 when considering the very low GFR levels. The differences between the two equations by weighted average analysis for strata of GFR >60 ml/min/1.73 m 2 were again clinically insignificant. CONCLUSION: POPULATION VERSUS PATIENT IMPLICATIONS At best, the equation might be considered as an evolution but not a revolution. By improving the systematic error of the equation in comparison with the one, the advantage of the new equation is when we think in terms of population [7, 14, 17, 65]. In this view, it seems that the equation performs better to define the CKD stage 3 prevalence in a general population, and especially in the younger subjects [66]. However, without the confirmatory mgfr test, this conclusion is not reliable and there are still reasons to think that the new equation overestimates the CKD prevalence in the Caucasian population [17]. In the same thought, it has been recently proved in different cohorts that the better bias linked to the equation leads to a better prediction of the associated risk of mortality [65, 67, 68]. Although impressive, we have already underlined the limitations of this type of epidemiological studies [69]. The superiority of the equation could be accepted in epidemiological trials but this is certainly more questionable if we are thinking in terms of the patient. If we have to know and follow the GFR of a given individual, the concept of precision becomes very important and in this context, the equation is clearly lacking any additional value. This assertion is reinforced if we consider the percentage of subjects who are correctly classified into the five CKD stages. In the validation data set, Levey et al. [7] demonstrated a concordance between the measured and estimated CKD stages of 69%, while Murata et al. [21] observed a correct classification of >70% only in 6 of the 25 GFR studied groups. Bjork et al. [32] found an overall proper CKD classification of 69%. All these results should be considered suboptimal as in the case of the 38% overall CKD misclassifications reported for the equation in a recent review [9]. Obviously, both the and equations are not magic. Serum creatinine remains the principal variable of the GFR-estimating equations, and if it is not accurately representing the individual s muscular mass [70], there is little chance that including the creatinine into the equations would improve their performance. Classical examples are hyperfiltrating diabetic patients [49, 59, 71] and cirrhotic [72, 73], renal transplanted [21, 51, 55], elderly [21, 73] and anorectic patients [73]. As clinicians, we have to know the GFR of individuals, not of populations. We have to treat patients, not statistical risks by associations. Therefore, we have to minimize the random error of the estimator, not only the systematic one. In fact, the true question might not be Is the equation better for the estimation of my given patient? but maybe Is there any chance that any creatinine-based equation is precise enough in my patient? and Would it not be better to measure GFR by a reference method in my specific patient? [20] Downloaded from P. Delanaye et al.

4 Table 1. Performance of and equations (using calibrated SCr values) in studies reporting accuracy data for both equations and with minimum of 50 measured GFRs Study GFR method SCr calibration Population N mgfrs Mean mgfr ± SD (range) Accuracy Bias Precision 30% 15% Mean Median SD of mean bias Murata et al.[21] Iothalamate Yes IDMS Mixed ± Levey et al. [7] 125 I-iothalamate, Iohexol, 99m Tc- DTPA Yes IDMS Mixed ± Eriksen et al.[39] Iohexol Yes IDMS General (no CKD) ± Bjork et al. [32] Iohexol Yes IDMS Mixed (12 116) Buron et al. [58] Inulin Yes LCMS KT recipients ± 18 (15 90) Nyman et al.[47] Iohexol Yes IDMS Mixed (9 121) Iliadis et al. [57] Yes IDMS DM Type ± Lane et al. [60] Cirillo et al. [56] 125 I-iothalamate Yes ClClin Pre- and postnephrectomy (median) (4 142) Inulin Yes IDMS Mixed ± Michels et al.[26] a 125 I-iothalamate Yes IDMS Mixed ± Tent et al. [50] 125 I-iothalamate Yes ClClin Prenephrectomy ± Estimating GFR using the Teo et al. [54] White et al. [46] Postnephrectomy Redal- Baigorri et al.[48] a Poge et al. [55] Jones et al. [63] Kukla et al. [51] ± Yes IDMS CKD ± Yes IDMS KT recipients ± Yes IDMS Oncology ± Yes IDMS KT recipients (12 83) Yes IDMS Evaluation of GFR (5 150) I-iothalamate Yes IDMS KT recipients ± KT recipients 1 year post-kt ± Continued POLAR VIEWS IN NEPHROLOGY

5 POLAR VIEWS IN NEPHROLOGY P. Delanaye et al. Table 1. Continued Study GFR method SCr calibration Population N mgfrs Mean mgfr ± SD (range) Accuracy Bias Precision 30% 15% Mean Median SD of mean bias 1400 Silveiro et al. [59] Orskov et al. [52] a Praditprnsilpa et al.[62] Soares et al.[53] Bargnoux et al. [64] Tent et al.[61] Gerhardt et al. [44] Camargo et al. [49] Van Deventer et al.[45] Yes IDMS DM Type ± Yes IDMS Polycystic kidney disease (7 118) Yes IDMS CKD ± Yes IDMS Healthy ± Yes IDMS KT recipients ± I- iothalamate Yes ClClin CKD ± CKD ± Yes IDMS Liver transplant (48 57) Yes IDMS DM Type ± Healthy ± Yes IDMS CKD 50 N/A Calculated average weighted values from available data in all studies b Calculated average weighted values from available data in all studies with analysis for strata of mgfr >60 ml/ min/1.73 m 2c, simplified (4 variables), re-expressed with calibrated serum creatinine, Modification of Diet in Renal Disease Study equation;, Chronic Kidney Disease Epidemiology Collaboration equation; GFR, glomerular filtration rate; SCr, serum creatinine calibrated to IDMS (isotope dilution mass spectroscopy), LCMS (liquid chromatography-mass spectrometry) or ClClin (Cleveland Clinic Laboratory); mgfr, measured GFR in ml/min/1.73 m 2 urinary clearance unless otherwise described; Accuracy, % of GFR estimates within ±30% and ±15% of mgfrs; Bias, estimated minus measured GFR in ml/min/1.73 m 2 (values for the mean bias in italic type were calculated if possible whenever not provided); Precision, one standard deviation (SD) of mean bias in ml/min/1.73 m 2 ;, 99m technetium-diethylenetriamine pentaacetic acid;, 51 chromium-ethylene diamine tetraacetic acid; CKD, chronic kidney disease; KT, kidney transplant; DM, diabetes mellitus. a Study design included some SCr measurements not exactly on the day of GFR measurement. b Data calculated for accuracy ±30%, accuracy ±15%, mean bias, median bias and precision from analysis of 18245, 690, 12303, 9484 and 3572 mgfrs, respectively. c Data calculated for accuracy ±30%, accuracy ±15%, mean bias, median bias and precision from analysis of 1950, 132, 1072, 1048 and 1000 mgfrs, respectively.

6 CONFLICT OF INTEREST STATEMENT None declared. REFERENCES 1. Smith HW. The Kidney: Structure and Function in Health and Disease. New York: Oxford University Press Inc, 1951; K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002; 39: S Levey AS, de Jong PE, Coresh J et al. The definition, classification, and prognosis of chronic kidney disease: a KDIGO Controversies Conference report. Kidney Int 2011; 80: Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976; 16: Levey AS, Bosch JP, Lewis JB et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130: Levey AS, Coresh J, Greene T et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med 2006; 145: Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150: Kemperman FA, Krediet RT, Arisz L. Formula-derived prediction of the glomerular filtration rate from creatinine concentration. Nephron 2002; 91: Botev R, Mallie JP, Wetzels JF et al. The clinician and estimation of glomerular filtration rate by creatinine-based formulas: current limitations and quo vadis. Clin J Am Soc Nephrol 2011; 6: Delanaye P, Mariat C, Maillard N et al. Are the creatinine-based equations accurate to estimate glomerular filtration rate in African American populations? Clin J Am Soc Nephrol 2011; 6: Delanaye P, Cohen EP. Formula-based estimates of the GFR: equations variable and uncertain. Nephron Clin Pract 2008; 110: c48 c Rule AD, Rodeheffer RJ, Larson TS et al. Limitations of estimating glomerular filtration rate from serum creatinine in the general population. Mayo Clin Proc 2006; 81: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: Earley A, Miskulin D, Lamb EJ et al. Estimating equations for glomerular filtration rate in the era of creatinine standardization: a systematic review. Ann Intern Med 2012; 156: Coresh J, Selvin E, Stevens LA et al. Prevalence of chronic kidney disease in the United States. JAMA 2007; 298: Delanaye P, Cavalier E, Krzesinski JM. Determining prevalence of chronic kidney disease using estimated glomerular filtration rate. JAMA 2008; 299: Delanaye P, Cavalier E, Mariat C et al. or study equations for estimating prevalence of stage 3 CKD in epidemiological studies: which difference? Is this difference relevant? BMC Nephrol 2010; 11: Pottel H, Martens F. Are egfr equations better than IDMStraceable serum creatinine in classifying chronic kidney disease? Scand J Clin Lab Invest 2009; 69: Mariat C, Alamartine E, Afiani A et al. Predicting glomerular filtration rate in kidney transplantation: are the K/DOQI guidelines applicable? Am J Transplant 2005; 5: Stevens LA, Levey AS. Measured GFR as a confirmatory test for estimated GFR. J Am Soc Nephrol 2009; 20: Murata K, Baumann NA, Saenger AK et al. Relative performance of the and equations for estimating glomerular filtration rate among patients with varied clinical presentations. Clin J Am Soc Nephrol 2011; 6: Botev R, Mallie JP, Couchoud C et al. Estimating glomerular filtration rate: Cockcroft-Gault and Modification of Diet in Renal Disease formulas compared to renal inulin clearance. Clin J Am Soc Nephrol 2009; 4: Chudleigh RA, Dunseath G, Evans W et al. How reliable is estimation of glomerular filtration rate at diagnosis of type 2 diabetes? Diabetes Care 2007; 30: Froissart M, Rossert J, Jacquot C et al. Predictive performance of the modification of diet in renal disease and Cockcroft-Gault equations for estimating renal function. J Am Soc Nephrol 2005; 16: Ma YC, Zuo L, Chen JH et al. Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. J Am Soc Nephrol 2006; 17: Michels WM, Grootendorst DC, Verduijn M et al. Performance of the Cockcroft-Gault,, and new formulas in relation to GFR, age, and body size. Clin J Am Soc Nephrol 2010; 5: Rigalleau V, Lasseur C, Raffaitin C et al. The Mayo Clinic quadratic equation improves the prediction of glomerular filtration rate in diabetic subjects. Nephrol Dial Transplant 2007; 22: Rostoker G, Andrivet P, Pham I et al. A modified Cockcroft- Gault formula taking into account the body surface area gives a more accurate estimation of the glomerular filtration rate. J Nephrol 2007; 20: Tidman M, Sjostrom P, Jones I. A comparison of GFR estimating formulae based upon s-cystatin C and s-creatinine and a combination of the two. Nephrol Dial Transplant 2008; 23: Urbaniak J, Weyde W, Smolska D et al. S-cystatin C formulae or combination of s-cystatin C and s-creatinine formulae do not improve prediction of GFR. Nephrol Dial Transplant 2008; 23: White C, Akbari A, Hussain N et al. Chronic kidney disease stage in renal transplantation classification using cystatin C and creatinine-based equations. Nephrol Dial Transplant 2007; 22: Bjork J, Jones I, Nyman U et al. Validation of the Lund-Malmo, chronic kidney disease epidemiology () and modification of diet in renal disease () equations to estimate glomerular filtration rate in a large Swedish clinical population. Scand J Urol Nephrol 2012; 46: POLAR VIEWS IN NEPHROLOGY 1401 Estimating GFR using the

7 POLAR VIEWS IN NEPHROLOGY 33. Rule AD, Bergstralh EJ, Slezak JM et al. Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 2006; 69: Rule AD, Larson TS, Bergstralh EJ et al. Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med 2004; 141: Stevens LA, Manzi J, Levey AS et al. Impact of creatinine calibration on performance of GFR estimating equations in a pooled individual patient database. Am J Kidney Dis 2007; 50: Panteghini M. Enzymatic assays for creatinine: time for action. Scand J Clin Lab Invest Suppl 2008; 241: Pieroni L, Delanaye P, Boutten A et al. A multicentric evaluation of IDMS-traceable creatinine enzymatic assays. Clin Chim Acta 2011; 412: Schwartz GJ, Schneider MF, Maier PS et al. Improved equations estimating GFR in children with chronic kidney disease using an immunonephelometric determination of cystatin C. Kidney Int 2012; 82: Eriksen BO, Mathisen UD, Melsom T et al. Cystatin C is not a better estimator of GFR than creatinine in the general population. Kidney Int 2010; 78: Kwong YT, Stevens LA, Selvin E et al. Imprecision of urinary iothalamate clearance as a gold-standard measure of GFR decreases the diagnostic accuracy of kidney function estimating equations. Am J Kidney Dis 2010; 56: Cavalier E, Rozet E, Dubois N et al. Performance of iohexol determination in serum and urine by HPLC: validation, risk and uncertainty assessment. Clin Chim Acta 2008; 396: Delanaye P, Cavalier E, Froissart M et al. Reproducibility of GFR measured by chromium-51-edta and iohexol. Nephrol Dial Transplant 2008; 23: Ng DK, Schwartz GJ, Jacobson LP et al. Universal GFR determination based on two time points during iohexol disappearance. Kidney Int 2011; 80: Gerhardt T, Poge U, Stoffel-Wagner B et al. Estimation of glomerular filtration rates after orthotopic liver transplantation: evaluation of cystatin C-based equations. Liver Transpl 2006; 12: van Deventer HE, George JA, Paiker JE et al. Estimating glomerular filtration rate in black South Africans by use of the modification of diet in renal disease and Cockcroft-Gault equations. Clin Chem 2008; 54: White CA, Akbari A, Doucette S et al. Estimating glomerular filtration rate in kidney transplantation: is the new chronic kidney disease epidemiology collaboration equation any better? Clin Chem 2010; 56: Nyman U, Grubb A, Sterner G et al. The and equations to estimate GFR. Validation in the Swedish Lund- Malmo study cohort. Scand J Clin Lab Invest 2011; 71: Redal-Baigorri B, Stokholm KH, Rasmussen K et al. Estimation of kidney function in cancer patients. Dan Med Bull 2011; 58: A Camargo EG, Soares AA, Detanico AB et al. The Chronic Kidney Disease Epidemiology Collaboration () equation is less accurate in patients with Type 2 diabetes when compared with healthy individuals. Diabet Med 2011; 28: Tent H, Rook M, Stevens LA et al. Renal function equations before and after living kidney donation: a within-individual comparison of performance at different levels of renal function. Clin J Am Soc Nephrol 2010; 5: Kukla A, El-Shahawi Y, Leister E et al. GFR-estimating models in kidney transplant recipients on a steroid-free regimen. Nephrol Dial Transplant 2010; 25: Orskov B, Borresen ML, Feldt-Rasmussen B et al. Estimating glomerular filtration rate using the new equation and other equations in patients with autosomal dominant polycystic kidney disease. Am J Nephrol 2010; 31: Soares AA, Eyff TF, Campani RB et al. Performance of the CKD Epidemiology Collaboration () and the Modification of Diet in Renal Disease () Study equations in healthy South Brazilians. Am J Kidney Dis 2010; 55: Teo BW, Xu H, Wang D et al. GFR estimating equations in a multiethnic Asian population. Am J Kidney Dis 2011; 58: Poge U, Gerhardt T, Stoffel-Wagner B et al. Validation of the formula in patients after renal transplantation. Nephrol Dial Transplant 2011; 26: Cirillo M, Lombardi C, Luciano MG et al. Estimation of GFR: a comparison of new and established equations. Am J Kidney Dis 2010; 56: Iliadis F, Didangelos T, Ntemka A et al. Glomerular filtration rate estimation in patients with type 2 diabetes: creatinine- or cystatin C-based equations? Diabetologia 2011; 54: Buron F, Hadj-Aissa A, Dubourg L et al. Estimating glomerular filtration rate in kidney transplant recipients: performance over time of four creatinine-based formulas. Transplantation 2011; 92: Silveiro SP, Araujo GN, Ferreira MN et al. Chronic Kidney Disease Epidemiology Collaboration () equation pronouncedly underestimates glomerular filtration rate in type 2 diabetes. Diabetes Care 2011; 34: Lane BR, Demirjian S, Weight CJ et al. Performance of the chronic kidney disease-epidemiology study equations for estimating glomerular filtration rate before and after nephrectomy. J Urol 2010; 183: Tent H, Waanders F, Krikken JA et al. Performance of study and equations for long-term follow-up of nondiabetic patients with chronic kidney disease. Nephrol Dial Transplant 2011; 2012 Suppl 3: iii Praditpornsilpa K, Townamchai N, Chaiwatanarat T et al. The need for robust validation for -based glomerular filtration rate estimation in various CKD populations. Nephrol Dial Transplant 2011; 26: Jones GR. Use of the equation for estimation of GFR in an Australian cohort. Pathology 2010; 42: Bargnoux AS, Servel AC, Pieroni L et al. Accuracy of GFR predictive equations in renal transplantation: validation of a new turbidimetric cystatin C assay on Architect c8000(r). Clin Biochem 2012; 45: White SL, Polkinghorne KR, Atkins RC et al. Comparison of the prevalence and mortality risk of CKD in Australia using the CKD Epidemiology Collaboration () and Modification of Diet in Renal Disease () Study GFR estimating equations: the AusDiab (Australian Diabetes, Obesity and Lifestyle) Study. Am J Kidney Dis 2010; 55: van den Brand JA, van Boekel GA, Willems HL et al. Introduction of the equation to estimate glomerular filtration 1402 Downloaded from P. Delanaye et al.

8 rate in a Caucasian population. Nephrol Dial Transplant 2011; 26: Matsushita K, Selvin E, Bash LD et al. Risk implications of the new CKD Epidemiology Collaboration () equation compared with the Study equation for estimated GFR: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Kidney Dis 2010; 55: Matsushita K, Mahmoodi BK, Woodward M et al. Comparison of risk prediction using the equation and the study equation for estimated glomerular filtration rate. JAMA 2012; 307: Delanaye P, Schaeffner E, Ebert N et al. Normal reference values for glomerular filtration rate: what do we really know? Nephrol Dial Transplant 2012; 27: Perrone RD, Madias NE, Levey AS. Serum creatinine as an index of renal function: new insights into old concepts. Clin Chem 1992; 38: Nair S, Hardy KJ, Wilding JP. The Chronic Kidney Disease Epidemiology Collaboration () formula performs worse than the Modification of Diet in Renal Disease () equation in estimating glomerular filtration rate in Type 2 diabetic chronic kidney disease. Diabet Med 2011; 28: Xirouchakis E, Marelli L, Cholongitas E et al. Comparison of cystatin C and creatinine-based glomerular filtration rate formulas with 51Cr-EDTA clearance in patients with cirrhosis. Clin J Am Soc Nephrol 2011; 6: Segarra A, de la Torre J, Ramos N et al. Assessing glomerular filtration rate in hospitalized patients: a comparison between and four cystatin C-based equations. Clin J Am Soc Nephrol 2011; 6: OPPONENT S COMMENTS We congratulate Dr Delanaye and colleagues on their scholarly review of GFR estimation and the comparative performance of the versus Study equations. We agree with them about the importance of ascertainment of GFR in clinical decisions, the practicality of estimating the GFR from serum creatinine, the greater accuracy of the versus study equation and the persisting imprecision of GFR estimates as an irremediable limitation of variation in non- GFR determinants of serum creatinine. We also agree that the equation represents an evolution, not a revolution, in GFR estimation, but we disagree with their conclusion that we should not make the change to report the egfr using the equation. At present, there are hundreds of millions of creatinine measurements performed each year for the purpose of estimating the GFR. For the vast majority of these measurements, it is not possible to measure GFR. Our disagreement rests on three main arguments. First, their discussion on the limitations in measurement methods is not relevant when comparing the performance of two equations in the same dataset. Welldone studies, such as ours [1], show a consistent improvement in bias at a higher GFR and in overall accuracy [2]. Second, they erroneously characterize the performance of the two equations as comparable. While we agree that the improvement in precision is small, with the improvement in bias, the resulting improvement in accuracy is clinically important. For example, in our study, the percent of large errors ( >30% of the measured GFR) decreased from 19.4 to 15.9% (18%) across the range of GFRs, and from 17.7 to 13.2% (25%) for egfr ml/min per 1.73 m 2 [1]. These results represent a large and meaningful improvement in performance. Third, they dismiss the reduction in prevalence estimates and improvement in risk stratification as not relevant to clinical practice. However, these improvements translate directly to more accurate individual decision making for disease detection and prognosis, key elements in patient care. Moreover, these improvements are observed in large subgroups at low risk in whom there has been concerns over overdiagnosis of CKD (middle-aged people, women and whites). Reporting egfr using the equation would mean that substantially fewer patients would have to worry about an erroneous diagnosis of kidney disease and the risk for the future. At this time, the equation maximizes the information available from the serum creatinine concentration. Evolution is a slow process. Gradual changes cannot be ignored. REFERENCES Lesley A. Inker and Andrew S. Levey 1. Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150(9): Earley A, Miskulin D, Lamb EJ et al. Estimating Equations for Glomerular Filtration Rate in the Era of Creatinine Standardization: A Systematic Review. Ann Intern Med Received for publication: ; Accepted in revised form: POLAR VIEWS IN NEPHROLOGY 1403 Estimating GFR using the